Microfluidics is a promising technology that is used extensively in biomedical devices, so called lab-on-a-chip devices. These devices harness a network of microchannels to mix, react, and conduct fluid flow. Most microchannel fabrication methods produce a stochastic surface roughness with heights ranging in the micro- to nano- scale. This inherent, stochastic roughness can potentially be harnessed to enhance microfluidic operations. Previous research on rough surfaces in microfluidics has focused on periodic, micro-scale obstructions, not of any stochastic nature. The purpose of this research is to characterize the effect of stochastic nano-scale surface roughness on microfluidic flow using very large-scale direct numerical simulations (DNS) and micro- particle image velocimetry (micro-PIV).

The two studies are focused on a microchannel with one of the walls, the bottom surface, which has a manufactured surface roughness using a hydrofluoric-acid (HF) etching process. The rough surface is scanned by an optical profilometer, and the exact topography is imported as the bottom surface of the computational microchannel. HF-acid etched glass and un-etched glass surfaces are directly compared to each other. In the first study, the DNS simulations are compared to micro-PIV experiments for a Newtonian fluid (water). The flow regime was laminar, diffusion dominated and limited to Re < 10.

The second study used a longer microchannel relative to the first study that was made possible by stitching together consecutive profilometer surface scans. This study only used simulations to study the effect of nano-scale roughness on microfluidic flow (with the previous study forming a basis for model validation). In the future, the study will be extended to Newtonian as well as non-Newtonian (shear-thinning) fluids in the same flow regime as the first study.

Overall, we have shown that an experimentally validated and experimentally driven three-dimensional computational study for microfluidic stochastic surface roughness is possible. Additionally, we have shown that the stochastic nature of the surface roughness and its effect on fluid flow can be characterized with numerous tools including velocity-perturbation contours, autocorrelation length (ACL), and energy spectra analysis.

The different analyses illustrated the effect of the rough surface in different ways. Velocity-perturbation contours showed that both the etched and un-etched rough surfaces produced very small velocity structures (eddies) very near the rough surface that merge to form larger structures as the height above the rough surface increases. The velocity-perturbation contours revealed an increase in the magnitude of the velocity perturbations by an order of magnitude by using the etched glass, which was directly caused by the increase in roughness height from HF etching. The ACL analyses also showed how the surface roughness produces small perturbation structures that merge and persist well into the midplane of the microchannel. Energy spectra analyses revealed a transfer of energy caused by the structures of the rough surfaces. Notably for the same Reynolds number, the etched surface produced velocity-perturbation structures that contained more energy and persisted higher into the microchannel compared to the un-etched surface.

This research has shown that a chemical etching surface treatment and other stochastic rough surfaces, even at the nano-scale, have an effect on microfluidic flow that can be characterized and potentially be harnessed across a range of fluid flow rates. Devices that use microchannels such as lab-on-a-chip medical devices can therefore be tuned and optimized for their respective applications such as reagent mixing, bubble creation and transport, fluid transport, and cell manipulation using stochastic surface roughness.